Study Of Steering Control Of Intelligent Vehicle For Curve Driving Based On Human-like Driver Model

Intelligent vehicle is the development direction of automobile industry.Steering control is an important content of intelligent vehicle motion control research.The curve is the key scenario of intelligent driving steering control function verification.Driver model,as a mathematic model to describe the human driver’s control behavior,plays an irreplaceable role in the development and application of intelligent vehicles.Therefore,making intelligent vehicles to exhibit the steering control behavior like human drivers in curve driving based on human-like driver model,which is of great significance for the development of intelligent vehicles and the improvement of the applicability,acceptability,and driving safety of intelligent driving systems.For this purpose,this paper is focus on the study of steering control of intelligent vehicle for curve driving,including collection of human driving data,establishment of human-like driver model,design of angle controller for electric power steering system,and comparison and analysis between real vehicle test and simulation test.The main contents are as follows:Firstly,perceptual behavior of human driving in curves and collection methods of driving data are analyzed.Based on driver-in-the-loop simulation platform,multicurvature test roads covering a large range of radius of curvature are designed,and driving simulation tests are carried out at various speeds.Human driving data in curve driving including vehicle motion states,road information,and driver steering signal are collected to provide training sample and evaluation basis for the establishment and verification of human-like driver model.Secondly,based on the analysis of visual attention mechanism of human driving on curves,a visual perception module is constructed to extract input parameters for steering decision.Considering the fuzzy reasoning behavior and complexity of human driving,a decision-making module of steering wheel angle is constructed based on datadriven approach of adaptive neuro-fuzzy inference system,and the mapping relationship between visual parameter inputs and steering wheel angle output is obtained.From the aspects of driving information perception and vehicle control decision,a human-like driver model is established to mimic the human steering maneuver.The simulation results show that the human-like driver model has a basic road following ability,the generated angle changes smoothly,can reflect the general human steering behavior in curve driving,and has certain applicability to other driving conditions.Then,the EPS system and vehicle dynamics model are established to provide the controlled object for controller design and simulation verification.The principle and system characteristics of automatic steering angle control are analyzed.A double closed-loop control strategy of motor current and steering angle was adopted,and the controller is designed based on PID and active disturbance rejection control algorithms.Inputting various motor current signals and steering angle signals,the inner-loop current tracking control and double closed-loop angle tracking control are simulated and verified respectively.The results show that the steering angle controller designed for the EPS system can track the steering angle accurately,quickly,and smoothly.Finally,combining the human-like driver model and the angle controller of EPS system,the steering control system based on human-like driver model and its simulation model are constructed.The real vehicle and simulation driving tests are carried out on actual urban roads respectively.The steering control data and steering control results are compared and analyzed.The results show that the human-like driver model based on the data of simulation driving tests can reflect the human steering behavior in real driving,and the automatic steering control based on human-like driver model can reproduce the driving trajectory of real curve driving.